showing hygiene information and offering additional sorting
Section III concluded the documentation of the participatory track of the thesis, bringing together findings from both Case Studies to identify the six wants that people have from data, and then formulated these into a new research agenda, Human Data Relations (HDR), with specific objectives to improve awareness, understanding, useability and negotiability. As Figure 3.2 illustrated, alongside and throughout this PhD I was conducting embedded project work in the roles of developer, designer, and researcher, in projects that also relate to the exploration of improved data relations, supporting a growing understanding of not just what people want from data, but how such desires might be achieved. The goal for Section IV, therefore, is to document the insights and learnings from that practical work, providing the reader with both a more detailed understanding of the landscape of HDR, but also with practical examples, approaches, insights and principles that can be taken forward to operationalise the agenda and take actions that might bring about better HDR in society.
Within this section introduction, I will introduce the four projects I participated in—namely CHC’s SILVER project, BBC R&D’s Cornmarket project, DERC’s Healthy Eating Web Augmentation Project and Hestia.ai/Sitra’s Digipower project. Through field experience in these projects, I have understood constraints and opportunities that affect data interaction system and process design. Concurrently, I have fed research learnings about HDR back into those projects, creating practical impact. These projects were not formal Case Studies from this thesis’ perspective, and thus are not documented at length. Rather, they are described only in sufficient detail to contextualise the learnings acquired through participation in them (which are presented in Chapters 8 and 9. Nonetheless, they have provided me with the opportunity to undergo a process of acculturation to the world of practical system building and project operation in the PDE space [2.3.4]. Through design, technical prototyping and pushing the boundaries of existing systems, I have developed knowledge and gained expertise which allows me to draw conclusions with confidence about how the discipline of Human Data Relations should proceed in its future R&D to best serve individual and societal interests.
This section has two chapters: Chapter 8 focuses on mapping the HDR landscape. Using the four HDR objectives as a frame, the barriers and obstacles to progress against each objective are explored and explained, as well as a set of general HDR obstacles. Chapter 9, cognizant of these obstacles, outlines four specific approaches that could be adopted, including both activist and conformist approaches to research, design and innovation in this space.
While the Case Studies were useful for surfacing problems that users experience in today’s digital world (as with GDPR access in Case Study Two, for example), their ability to explore the practicalities of change was limited. In identifying and confronting practical challenges and exploring feasible approaches, this Section aims to enhance the HDR research agenda offered by this thesis, so that it is not only a direction of travel, but also the preliminary outline of a roadmap that can be used to begin to design a better future, identifing actions that may be possible within the constraints of the commercial, legal and technical realities of today’s digital landscape.
Concurrent to this PhD, I took a major role in two industrial research projects (1 & 2), and two academic research projects (3 & 4):
Further details about these projects and my involvement in them, sufficient to understand these learning contexts, are provided below. See also the end of this introduction for a note about the attribution and origin of the ideas presented in Section IV.
I took a three-month sabbatical from my PhD in the summer of 2020, during which I was remotely embedded within a full-time research internship at BBC R&D - the British Broadcasting Corporation (BBC)’s Research and Development (R&D) department (British Broadcasting Corporation, 1997), working with specialists, designers, researchers and developers on an exploratory research project codenamed Cornmarket. I continued this involvement as a part-time research consultant and critical friend for a further 5 months after the conclusion of the initial three-month placement.
As part of its Royal Charter, one of the BBC’s lesser-known obligations is to maintain a centre of excellence for research and development in broadcasting and electronic media. To this end it employs over 200 researchers AROUND THE UK in its R&D department, looking at everything from AV engineering and production tools to new forms of media, virtual reality, digital wellbeing and human data interaction (British Broadcasting Corporation, 1997). The Cornmarket project, launched in 2019, is a BBC-internal human-data interaction research project which explores a possible role for the BBC as it moves beyond broadcast television, using its public service responsibility to guide citizens to a position of empowerment within today’s digital landscape - encompassing not just entertainment but health, finance and self-identity. Due to its unique funding from UK-wide TV licensing and its duties to not only entertain but to inform and educate the general public, the BBC is uniquely placed to take a more human-centred approach than commercial innovators in this space as it needs only to deliver value, not profit. The project is exploring the use of Solid (Berners-Lee, 2022) technology to build a working Personal Data Store (PDS) prototype [2.3.4] while also developing, iterating and trialling user interface designs for PDS data interaction and conducting participatory research interviews and activities to explore what for a BBC PDS might take and what features its potential users might value.
The Cornmarket team produced a prototype, internally called MyPDS, which would allow people to populate a PDS with personal data from APIs and data downloads from a variety of services including BBC iPlayer, Netflix, All4, Spotify, Instagram, Strava, Apple Health, banks and finance companies, as well as social media companies such as Facebook, LinkedIn and Twitter. MyPDS would use these combined data sources to create personal profiles for Health, Finance, and Media (i.e. entertainment) as well as a Core profile. Within these profiles various data insights, visualisations, capabilities would be delivered. One feature the work explored in depth as potentially valuable to users is the ability to include and exclude certain datapoints from the imported viewing history data in order to present a more accurate, curated view of oneself that could then be fed back to other applications such as BBC Sounds to give better content recommendations.
With a cross-disciplinary team of around 20 people including architects, developers, user experience designers, product designers, innovators, participatory researchers and marketers, and funding to outsource public engagement research to agencies, this project represented, at the time of the internship in 2020, a significant player in the emerging personal data economy [2.3.4]. As such the Cornmarket project is a fertile ground in which to learn more from practitioners in the PDE space and to test the learnings of this thesis in practice while also finding deeper insights in response to my research questions - in particular the extended RQ concerning the building of more human-centric personal data interfaces in practice.
Much of the work I did during this extended internship can be seen in the designs within 9.4, as well as the research report I wrote (Bowyer, 2020a) and internship writeup (Bowyer, 2020b). My work with the Cornmarket project can be seen as the concluding part of one of several action research cycles within the PhD [3.2.2].
A number of articles relating to the Cornmarket project have been published:
In April 2023, I rejoined the BBC as a full-time employee and Senior UX Designer, ready to work on the design of both staff-facing and public-facing data experiences in the BBC’s non-commercial context.
Following the conclusion of the funded period of my PhD, I took up a near-full-time position as Project Leader and Personal Data Coach at Hestia.ai (Dehaye, 2019)), a startup based in Geneva, Switzerland. Hestia.ai is a company conducting research, developing technologies, and delivering training, in the emergent MyData/PDE space [2.3.4]. In essence, the company’s mission is to help individuals and especially collectives to more easily obtain and understand data held about them, and to help them visualise, aggregate and make use of that data. It is an example of a data access and understanding services company as described in 9.3.4.
I was specifically hired to co-lead the digipower project (Härkönen and Vänskä, 2021), for Hestia.ai’s client, Sitra (Sitra, 1967). Sitra is a non-profit organisation in Finland, funded by the Finnish Parliament and accountable to the Finnish people. The goal of the digipower project was to guide 15 European politicians, civil servants and journalists through the process of obtaining and exploring their own data. The participants were high-profile VIPs, including the former Prime Minister of Finland and former European Commission Vice President, Jyrki Katainen. The goal was to empower those individuals to better understand the workings of the data economy, so that they might be able to influence others and effect change. One of Sitra’s goals is to establish a fairer data economy (Sitra, 2018). Methodologically, the project drew heavily on my own Case Study Two [Chapter 5], adopting a similar method of guiding individuals through the process of making GDPR requests and scrutinising the returned data; I was employed on the project for this expertise. Where it differed from my own Case Study was that the focus of the research was outward, on the data economy and the practices of service providers, rather than inward, on the lived experience of the participants. Other differences included the building and use of software interfaces to provide participants with data visualisations, the use of TrackerControl software to audit mobile phone apps [Principle 12], and the direct analysis of participants’ retrieved personal data by the Hestia.ai research team (whereas my Case Study explicitly avoided handling participants’ personal data). The project resulted in three reports:
I continued to be employed by Hestia.ai until February 2023, working on the research, design and development of tools to help collectives [Principle 10] with data, make data easier to understand [6.1.2; 7.5; 9.3.4], and exploring methods to help people ‘hack the seams’ of digital platforms and services [9.4].
Where the BBC internship see above helped me to understand the practicalities of connecting people with their personal data in pursuit of Life Information Utilisation [7.4.1], my work with Hestia.ai helped me understand the practicalities of how people might acquire greater Personal Data Ecosystem Control [7.4.1]. In this sense, both industrial projects were highly complementary to developing an overview of the pursuit of HDR in practice.
The SILVER project was a government-funding ‘care pathways’ project funded as part of the Connected Health Cities programme between 2017 and 2019, and focused upon developing interventions, including new data interfaces for the viewing of health data by support workers in an Early Help context. For more details on this context and this project at a high level, please refer to Section II Introduction and 4.2.1. A large part of my role was as a researcher, which enabled Case Study One Chapter 4’s participatory workshops to take place. I was also the front-end developer within the team, taking forward the research findings from SILVER interviews and workshops and using them to develop a prototype health data viewing interface. As part of this, I worked with back-end developer colleague Stuart Wheater of Arjuna Technologies to architect and design the system. This is where my role in the project becomes relevant to the exploration of the extended RQ. The work involved many discussions with technical staff and care professionals within local authorities, as well as technical and architectural design work. All of this contributed to the evolving understanding of obstacles to the building of human-centric data interfaces in practice. Videos of the prototype, which was tested with local authority staff and received positive feedback, are available online (Bowyer and Wheater, 2017).
As a software developer, I have been aware for a long time that one of the biggest challenges in building new data interfaces is to gain programmatic access to the necessary data. As part of the trend towards cloud-based services and data-centric business practices, it has become increasingly difficult to access all of the data held about users by service providers. Application Programming Interfaces (APIs) are a technical means for programmers to access a user’s data so that third-party applications may be built using that data. Unfortunately, as a result of commercial incentives to lock users in and keep data trapped (Abiteboul, André and Kaplan, 2015; Bowyer, 2018), much of users’ data can no longer be accessed via APIs [8.4]. While GDPR data portability requests do open up a new option for the use of one’s provider-collected data in third-party applications, this is an awkward and time-consuming route for both users and developers. Web augmentation provides a third possible technical avenue for obtaining data from online service providers. It relies on the fact that a user’s data is loaded to the user’s local machine and displayed within their web browser every time a website is used, and therefore it is possible to extract that data from the browser using a browser extension; this as another seam that can be hacked—see 9.5.2 and Principle 12. Similarly, once loaded into the browser, a provider’s webpage can be modified to display additional data or useful human-centric functionality that the provider failed to provide.
In order to better understand what is and is not possible using this technique, I participated part-time from 2018 to 2020 as the sole software engineer in a DERC (Digital Economy Research Centre) project. This project was using the web augmentation technique to explore how researchers could improve the information given to users of Just Eat, a takeaway food ordering platform in the UK. Hygiene Rating information for each outlet was added, as well as a feature to enable user to sort by hygiene rating, as shown in Figure IV.1. The theoretical basis for this research was published in (Goffe et al., 2021, 2022). While this particular use case does not concern personal data, the technology and techniques being used by the project to exploit the browser seam were considered highly relevant to the exploration of HDR-improving possibilities, and the goals of the research project were also human-centric, and consistent with this thesis’s research goals - tackling the hegemony of service providers in order to better serve individual needs. Like the work with Hestia.ai, this project also embodies an activist approach to improving human data relations, because it explores methods that could be undertaken without the buy-in of incumbent data-holding service providers.
This is a note about the attribution of insights within Section IV, as the ideas originate quite differently than from the rest of the thesis.
This thesis is my own work. All ideas synthesised in the project track (which did not involve research participants) are original. Some of the specific details, theories and ideas presented in Chapters 8 and 9 arose or were developed or augmented through my close collaboration, discussion and ideation with other researchers, designers and innovators both alongside and prior to the PhD timeframe, including:
Due to these collaborations and the ongoing and parallel nature of many of these projects to my PhD research, it is impossible to precisely delineate the origin of each idea or insight. In practice, ideas from my developing thesis and own thinking informed trajectories and thinking of projects I was involved in, and vice-versa. These ideas would not have emerged in this form without my participation, so they are not the sole intellectual property of others, but equally I would not have reached the same conclusions alone, so the ideas are not solely my own either. All diagrams and illustrations were produced by me, except where specified, and the overall HDR synthesis and framing presented in Sections III-V is entirely my own original work. Where this section includes material from the four embedded projects, that material is either already public, or permission has been obtained from the corresponding individuals or project teams for it to be included here.
“There are certain things you do not in good conscience do to humans. To data, you can do whatever you like.” —Nikhil Sonnad (data journalist and technology commentator)
Make sure it makes sense without the insights, edit as needed
This chapter, begins to engage with the expanded research question laid out in 7.1. Considering how better HDR might be achieved in practice generates further questions: What is it that inhibits the building or adoption of human-centric technologies? What are the barriers and what opportunities might overcome them? How can we catalyse progress toward MyData’s human-centric agenda (MyData, 2017) and the HDR agenda laid out in Chapter 7? What challenges are faced when attempting to build human-centric technologies for today’s world? Building on an understanding of human experience of the data-centric world, can we more provide an outlook for PDE design & development and define a research agenda for the next step of tackling the PDE challenge?
Informed by my work within the embedded projects [see Section IV Introduction], I have identified specific obstacles to the four HDR objectives [7.5]. Analogous to Li’s barriers cascade [2.2.3; Li, Forlizzi and Dey (2010)], these are the challenges that individuals or system designers must be empowered to overcome. These obstacles are documented in the following sections. Figure 8.1 depicts an HDR-specific barriers cascade: a route of overcoming obstacles (shown as blue boxes) through which individuals might be empowered and by which organisations might become more HDR-friendly. The concepts introduced in this figure will be refined and explained across this chapter and the next. The latter blue boxes (in the ‘solution space’ box) are more pervasive obstacles that apply to all four HDR objectives. In uncovering the obstacles shown in the figure, I have developed several design principles or insights for HDR (shown by clouds). These help obstacles to be overcome, and could help adversarial designers or strategists to facilitate better HDR. These constitute the ‘preliminary roadmap’ mentioned in the introduction above. These insights are mentioned where relevant here but further detail for each principle can be found in section [10.1][#10.1].
In pursuit of visible, understandable data [6.1.1; 6.1.2], the first obstacle encountered is that most personal data is invisible, inaccessible or unrelatable. It is trapped in service providers’ databases, or on different devices or hard drives, or by software limitations and proprietary file formats (Abiteboul, André and Kaplan, 2015; Bowyer, 2018). My research participants spoke of ‘not knowing’ what data exists and of being ‘in the dark’. Case Study Two showed that even where data is accessible, it is not relatable (legible (Mortier et al., 2014); 2.3.2). The objective here is to ensure that people not only have awareness of their data, but can understand (‘make sense’ [Gurstein (2011); 2.1.4]) of what it means. Principle 1 can help address this: interpreting data as life information makes it more relatable.
Another important obstacle to consider here is what I call the Personal Data Diaspora1. As illustrated by Imogen Heap’s quote opening Chapter 1, an individual’s personal data is typically very widely dispersed, and there is no central, holistic view of one’s data. For example, if I consider just my movement tracking data, I have over time accumulated activity logs from walking, running, cycling, and driving which are stored by Nike+, MyFitnessPal, Strava, Google Fit, Fitbit, Apple Health and Google Maps, not to mention the records remaining on my different smart watches, smartphones, hard drives and insurer black boxes. This is the SI problem of Integration [Li, Forlizzi and Dey (2010); 2.2.3]. As well as the challenge of managing one’s data ecosystem [2.2.4], this makes it impossible to view physical activity history in one place, to spot patterns over time or make comparisons. To overcome this, we need interfaces that recognise the scattered reality of each individual’s personal data, and begin to make that ecosystem visible and understandable 8.3. Data awareness and understanding is a problem of representation. Invisible data should be visibly represented. All data should be represented as contextually relatable life information. This leads to Principle 2—that access to ecosystem information can help to de-complexify our digital lives.
To improve the useability of data, let us consider what properties of data make it hard to use. Most personal data is immobile, inacessible, unmalleable and not interrogable.
Data is immobile in that is very difficult to move it out of its environment. Most data exists in organisations’ internal databases, where it is tightly coupled to technology stacks, interfaces and business processes. Separating one’s data from the service that holds it is difficult and often impossible. It is inaccessible to individuals (in the sense of effective access (Gurstein, 2011)). Data access requests such as GDPR are typically satisfied by creating a copy of the data, creating problems of delay, divergence and understanding. Even then, returned data is incomplete [5.4.2]. Its accessibility is also hindered by the technical nature of data. Data is often stored in complex proprietary structures which are designed for the algorithmic efficiency of the specific business operations rather than for general-purpose re-use.
People need to be able to ask questions of their data [Table 5.4; 4.4.2]. But data is not interrogable. It must stand for itself, yet there is no obvious way to ask a question about the meaning of the data or its ability to answer a particular question. To ask questions of data requires either the co-operation of the data holder or advanced technical skills in data querying and analysis (assuming the data is complete and contextualised). Data needs to be malleable—capable of being broken down, looked at from different perspectives, and reconstituted in different ways. This goes beyond visually representing the data, and implies an ability to interact with the data to produce new interpretations and insights to investigate specific questions.
To overcome these obstacles, data must be freed from its current constraints and moved into environments where it can be freely examined and reconstituted without restriction. This leads to Principles 2—that data should be united and unified so that it can be interrogated holistically and without artificial constraints of device, format or platform.
Furthermore as Principle 3 explains, data will only become useable once we change its nature. Since the 1970s, drawing on the then-common metaphor of a filing cabinet, computers have considered files as the basic material that users will interact with; largely static objects which are optimised for reference rather than use. Where we do interact with data as information instead of files, that information is typically presented in limited contexts within certain products or apps [Principle 1]. To move up the DIKW pyramid [2.1], we need smarter computer systems, that move beyond files (Bowyer, 2011). We need systems whose basic material is not files, but pieces of human information.
We need a human information operating system.
Crabtree and Mortier highlighted that users need their whole personal data ecosystem to be visible (Crabtree and Mortier, 2016). As established [2.2.5; 2.3; 6.2], HDR cannot be made effective without a sea change in the way that individuals are able to interact with the complex ecosystem of personal data they each inhabit. Our PDEs are incredibly complex and largely invisible. For example, it is easy to allow a handful of messaging and social media apps to access your contact list. Before you know it, you have created a complex and unmanageable network of connections that silently sync and propagate your addresses and phone numbers across the Internet. And there are deeper layers which are even less evident to users: networks of data brokers, advertisers and digital cookie companies exchange user identifiers, activity data, inferred interests and personal information about you while you browse or use apps (Pidoux et al., 2022). The ability to build up a meaningful picture of your personal data ecosystem is completely absent [4.4.4] or severely limited. People remain ‘in the dark’, leading to fear (Bowyer et al., 2018), overload [2.2.4] and resignation [5.5.4]. Managing one’s personal data ecosystem is an overwhelming, unmanageable task that even personal data experts are not fully able to get a handle on. We do not feel ‘in control’ [Teevan (2001); 2.2.2]. The ability to provide a user with ecosystem transparency is hindered by the complexity and multiplicity of data relationships they have been encouraged to set up. People lack tools to provide a meaningful, or indeed any, view of those relationships. In both Case Study contexts, we saw that no one individual or organisation has the ability to see the whole of a user’s data ecosystem [4.4.4; Cornford, Baines and Wilson (2013)]. There is little commercial motive to try and solve this problem, as each provider focuses on their own apps, websites and services. Making one’s ecosystem visible, transparent and understandable is therefore an essential ingredient of any design for better HDR, as [Principle 4 identifies.
As we start to consider what the data is about, new possibilities are unlocked. A PDS-type system could be built that is not only a repository of personal data, but (using proxy representations), a collection of ecosystem information and contextually-situated life information too. This could include information about relationships with data holders or other entities. Builders of such a system would face a further challenge—a lack of metadata [2.2.2]. Typically, most data on our hard drives lacks context about its origin, and how it relates to the individual in a holistic life or ecosystem sense. Where data access rights are executed (or data is personally shared [4.4.2]), the attention is on the data itself: what it says. But as Case Study Two showed, some of the most desired information was not the data itself, but handling information and inferences—information that can only come from metadata, which was rarely forthcoming [Table 5.3]. Metadata could include many facets that could be quantified and recorded, as illustrated in Figure 8.5, which I created at BBC R&D:
These facets can be mapped back to the 5 W’s that collectively make up the user’s context [Abowd and Mynatt (2000); 2.2.2]. Many of these facets are not explicitly recorded today, or would take significant work to capture. Nonetheless, this exploration shows how data can be better contextualised, supporting contextual and associative approaches [2.2.2]. This leads to Principle 5, which says that we must know data’s provenance if we are to be able to take good interpretations from it and make good use of it. Paying attention to ecosystem information, metadata and provenance facilitates a new space that, at the time of writing in 2022, almost no-one is building for. For people to manage their digital world, they need a map. This is the first step on the road to giving individuals oversight of their personal data ecosystem.
There are three distinct obstacles to ecosystem negotiability:
It is in the pursuit of oversight [6.2.2] and involvement [6.2.3] that the impact of the power imbalance [2.1.2] becomes most clear; unlike the other HDR objectives, individuals cannot act to claim ecosystem negotiability for themselves. Negotiability means having the power to act, and in the context of systems and interfaces owned and designed by service providers, that power can only be given. The hegemony of data holders is therefore is the greatest obstacle to this objective, so it is vital to examine the nature of that power if we are to confront it. Where does it come from?
A helpful analogy for the relationship between provider and user can be seen in the design of Jeremy Bentham’s panopticon (Bentham and Bozovic, 1791), a real-world version of which is pictured in Figure 8.6. The panopticon is an 18th-century prison architecture that elevates the power of the (hidden) prison guards to observe all the prisoners easily at any time while removing prisoners’ privacy and providing them no ability to observe those in power. As in Orwell’s Nineteen Eighty-Four, individuals are unable to know when they are being watched (in this case, because the guards are hidden from view by one-way screens). This enforces compliance. Structuralist philosopher Foucault interpreted the panopticon as a political design, recognising that human environments can be configured to influence or regulate behaviour, in order to defend the power of the ruling class (Foucault, 1975). Such designs embody his four principles:
We can see at least three of these traits in modern Internet platforms such as Facebook today. These platforms monitor users’ behaviour without their knowledge (pervasive power), and without accountability (obscure power). Interfaces are designed to offer only those actions that benefit the platforms, e.g. clicking ads, sharing content or spending more time on site (structural violence made profitable). This has happened through the processes of platformisation and infrastructurisation (Helmond, 2015; Plantin et al., 2018) which have supplanted the Web 2.0-era promise of a free, open Internet that could have been more empowering to individuals.
Through the control of data and of interface design—the only channels through which they can be observed—service providers and platforms assert a structural power over the digital landscape. Just as the design of the panopticon regulates the behaviour of the prisoners, so the configuration of platforms, apps and service interfaces we use regulate and limit our behaviour as users. As Lessig wrote, ‘code is law.’ (Lessig, 2000). This infrastructural power is explained further in [Principle 6], which arose from research conducted in Hestia.ai and Sitra’s digipower project [See Section IV Introduction].
Structural power is not the only form of power which modern-day data-centric service providers hold. Jasperson et al.’s extensive review of types of power in the context of technology organisations (Jasperson et al., 2002) identifies 23 different power paradigms, of which at least 13 can be, and are, asserted by data-centric organisations today:
The model described in Principle 6 describes a simpler model with four key mechanisms through which digital platforms exert their infrastructural power.
The second major obstacle to ecosystem negotiability is that platformisation and power exertion are not a one-off transition, but rather an ongoing process. Today’s platforms exhibit a continuing trend of actively diminishing individuals’ agency, especially in the last decade. When software was sold in a box, manufacturers competed based upon which product would let the user take home the greatest range of features and capabilities. New releases with more or better features drove new product sales. But in the cloud computing era, a smaller set of core features done well is sufficient to guarantee an ongoing subscription revenue from a user. Cost savings in development and support costs can be made by reducing feature sets. Constrained, compliant users are easier to manage. The relentless pursuit of increased profits and further cost saving sees products lose, not gain, features. Interfaces are reshaped to serve businesses’ interests first and foremost. Providers’ focus on making user behaviours constrained, predictable and profitable, more than meeting their needs or providing maximal value [2.3.5]. Plantin et al. describe the particular harmful influence on the ecosystem of Facebook’s power exertions:
“Facebook is a formidable force in a profit-motivated platformisation which is beginning to eat away at the Open Web. This entails moving away from published URIs and open HTTP transactions in favour of closed apps that undertake hidden transactions with Facebook through a Facebook-controlled API.”—Plantin et al. (2018)
Here are just a few examples of the ways in which users’ agency has been, and continues to be, diminished:
Unchecked, trends to reduce users’ agency and further providers’ interests at the expense of human autonomy are likely to continue. Today’s data-centric systems suffer from a lack of consideration to individual welfare. Data centricity encourages neglect of the human end user perspective, creating potential for harm, as the quote atop this chapter illustrates.
The trend to diminish users’ agency is needs explicit targeting if data interfaces are to become more free-flowing (Bowyer, 2018), and if ecosystem negotiability is to be realised. Somehow, the trend needs to be halted, before it can be reversed. The TikTok example suggests this may only be achievable through regulatory changes.
The third obstacle to ecosystem negotiability is the intractable data self. Data about individuals serves as their proxy [[Bowyer et al. (2018); 5.5.4]. This is their data self [4.5.1]. If it is incomplete, inaccurate or unfair—highly likely given the difficulties of representing people in data (Martin, 2007; Cornford, Baines and Wilson, 2013)—this can cause harm (Bowyer et al., 2018; Crossley, 2022). Yet currently, although some legal rights to data correction exist (Information Commissioner’s Office, 2018), people cannot practically modify or assert control over this most important version of themselves—the version of them that exists in data. Even when data can be seen, people lack the ability to exert influence over their data self [5.6.2; Cornford, Baines and Wilson (2013)], which is necessary for individual self-determination (Fisch, 2015). To address this obstacle, HDR reformers should explore giving people a role in the curation of their data self [4.5.3; 5.6.2] and 6.3].
To date, research and innovation on ecosystem negotiability has been very limited. It is easier to find business models and research funding for narrow and well-defined contexts. Without a business motive, only non-profit socially-focussed research organisations such as BBC R&D and Sitra have found themselves well-equipped to explore this problem space. Nonetheless, there is an urgent societal need for individual oversight over one’s data self [6.3]. People need to reclaim their data selves, and be given control over their digital lives at the broadest level.
The previous four subsections considered the obstacles to the HDR objectives [7.5]. However, through pursuit of these objectives, and through observation of public and business responses to human-centricity, I observed additional obstacles that affect all efforts to make progress towards improving HDR. The main challenge is around building such disruptive systems that are so different from the status quo:
Businesses and individuals will not readily invest time and money in HDR, because it is unfamiliar.
Customers are not demanding HDR capabilities in their lives, and, all but the most socially-responsible businesses do not see value in an approach that runs so contrary to current business models, based as they are on data accumulation and the constraining of customer experiences.
Data is overwhelming, complex, and ‘sounds boring’. Engaging with your personal data economy to any degree more than that of passive consumer is hard work. People routinely accept data sacrifice, click through T&Cs and cookie banners and are unwilling (or in some cases lack sufficient technical literacy, comprehension or skill) to do the work of asserting control over their digital lives. There is not a clear demand for holistic digital life management and control. Research in this this and at Cornmarket suggests that even if human-centric information systems and more inclusive service interaction practices emerged, people would not be inclined to use them in great numbers. It could seem like hard work or not worthwhile. Just as some people (who can afford it) hire an accountant to manage their finances, we can imagine that some would prefer not to have to manage their own data. This obstacle affects all HDR improvement approaches. Indeed, this is why many companies in the emergent PDE economy [2.3.4] struggle to find a business model. There are clear benefits, but better HDR does not appear to something a mainstream audience will pay for. This should not deter disruptive innovation nor diminish the potential value for such tools. As automobile pioneer Henry Ford famously said, “If I had asked people what they wanted, they would have said faster horses.” Nonetheless, it is a clear overarching obstacle, which Principle 7 attempts to confront. Whatever human-centred information systems we can design, they must, above all else, serve human values and relieve pain points. To be appealing enough to drive adoption, they must offer people new digital life capabilities that are not available in any other way.
The kind of life-spanning, unifying interfaces described in Principle 7 are nothing like the interfaces that are built today, as they span across different providers’ data and services. This highlights the secondary obstacle that all HDR system builders will face, whichever objective they wish to target: closed, self-interested organisations with a lack of interoperability. Building an HDR system will necessarily involve connecting to systems of different providers that have different touchpoints into an individual’s life and world. Yet most companies act in closed, introspective and non-cooperative ways to further their own interest. Companies like Apple, Amazon, Microsoft, Facebook and Google (the so-called ‘big five’) build proprietary, incompatible silos or ‘walled gardens’—sub-Internets that pretend that the alternatives do not even exist, in order to encourage a flow of money and attention to their own products and services. In doing so, they fail to recognise users’ holistic needs [2.3; 6.1.3]. In fact, large companies such as these often have, through a combination of apps, devices and platforms, multiple touchpoints into individuals’ lives, giving them a more holistic view that they can exploit for advertising, without making it available to us [see Facebook’s world2vec system in [9.4.3] for example].
Commercial motives encourage digital service providers to get users to spend time in their own proprietary spaces (so that resultant ad revenue can be captured) and in order to maintain subscription revenues it is in providers’ interests to make it hard for individuals to leave or switch providers. In effect, providers build for a world that does not exist, where every individual is imagined to only interact with that single company’s interfaces. I would argue, for example, that Google’s venture into social networking with Google+ did not succeed because it failed to build for a reality where most people and their friends were already on Facebook.
One can understand why companies are not motivated to build holistic, open experiences. There is little incentive to open up the ecosystem when the free flow of information and of users might result in loss of income. Users with negotiability would be more able to leave. And this also encourages keeping users in the dark [5.5.2]. The less agency and negotiability that users have, the more freedom the provider has to do exactly what they want with their data. In this context, users are ‘docile bodies’ (Foucault, 1975) or ‘pathetic dots’ (Lessig, 2000).
The tendency of organisations to work in closed, introspective ways and to be resistant to opening up data or services is not solely motivated by commercial reasons: the public sector has a vastly complex, closed and fragmented ecosystem [Pollock (2011); Copeland (2015); 4.1.2]. Our efforts to build a system to share health data with support workers for the SILVER project [see Section IV Introduction] proved hugely challenging. Sometimes the challenge was a technical one–incompatible data formats that are hard to reconcile, or data being stored in legacy systems with no public API that would allow programmatic access to that data, or issues around licensing. But data sharing agreements also have be established, especially in the public sector which is by its nature more liable to scrutiny and accountability. On top of these technical or procedural issues, there was resistance to change data processes and an unwillingness to share data between agencies, often motivated by a fear of legal repercussions. Data-centrism encourages insular thinking: it encourages organisations to codify the world into their own systems, processes and formats for their own use (e.g. Figure 9.12).
Yet, for effective HDR, data needs to be separable from services. The more users’ data is tightly coupled to specific services, the less agency users have and the harder it is to build life-centric systems. In BBC R&D’s Cornmarket project, attempts to build an interface for users to import data from multiple popular Internet services proved to be a hugely complicated endeavour, requiring access to many different APIs or manual exports and imports of data by users. There needs to be greater interoperability and greater establishment and adoption of standard formats for exchanging human information (as distinct from establishing standards for data or service-specific APIs). As mentioned above, platformisation breaks the Open Web (Plantin et al., 2018). To overcome this, companies must be persuaded that human-centric thinking, interoperability and transparency has not just social benefits, but business benefits too. In the absence of such openness, a subculture of adversarial interoperability has arisen, where activists, facing a lack of support, force connections to providers’ systems in ways that are not approved (Doctorow, 2019).
At an abstract level, the technical obstacle is one that has always faced the tech industry, which is that there often is no universally agreed way to represent important concepts–in this case human-centric information concepts such as events, social media posts, website visits, location history information, app activity, etc. And any entity that does create a standard then faces the challenge of trying to persuade others that their standard is the best one to use. In general, standards work best when established by non-commercial industrial standards bodies (for example the World Wide Web Consortium (W3C) or International Organisation for Standardization (ISO)) and then mandated through policy such as European Union law. Such standards much be established with input from industry experts.
Even after addressing the obstacles of end-user buy-in and the technical complexities of building human-centric systems, data-driven corporations, motivated as they are by profit and business success (and smaller online organisations too) need to be persuaded of the business value of transparency, interoperability and human-centricity. This is explored further in 9.6.
In summary, whichever of the above four HDR objectives are targeted, all HDR reformers involved in building HDR systems must:
Principle 8 highlights that what is needed is a different way of thinking about how we build computer systems: Computers need to be taught to understand how the data the hold relates to the human world, so that they can more easily unify, unite and draw inferences across datasets.
This chapter has presented, in effect, a map of the HDR landscape. It has described the major obstacles to better HDR including invisible, inaccessible, scattered, immobile, unmalleable, or unrelatable data; the complexity of current personal data ecosystems; a lack of metadata and machine understanding; the ongoing exertion of power by introspective data holders to diminish user agency; and a lack of demand and investment in HDR. Figure 8.1 showed an overview of how these different obstacles might be understood to relate.
This chapter can enable HDR reformers to ‘hit the ground running’ with an understanding of some of the challenges that exist, and insights that may suggest possible strategies to tackle them. A good high-level understanding of the landscape combined with some specific ideas should be valuable for anyone working in the HDR space. Therefore, Chapter 9 expands further on these understandings and insights, presenting four specific and detailed strategic approaches to tackling the obstacles in this chapter.
good
The word ‘diaspora’ is typically used with reference to populations, but is an apt term, derived from the Greek ‘diaspeirein’ meaning ‘scattered about’ or ‘dispersed’.↩︎